Padua transform

This demonstrates the Padua transform and inverse transform, explaining precisely the normalization and points

using FastTransforms

We define the Padua points and extract Cartesian components:

N = 15
pts = paduapoints(N)
x = pts[:,1]
y = pts[:,2];
nothing #hide

We take the Padua transform of the function:

f = (x,y) -> exp(x + cos(y))
f̌ = paduatransform(f.(x , y));
nothing #hide

and use the coefficients to create an approximation to the function $f$:

f̃ = (x,y) -> begin
    j = 1
    ret = 0.0
    for n in 0:N, k in 0:n
        ret += f̌[j]*cos((n-k)*acos(x)) * cos(k*acos(y))
        j += 1
#3 (generic function with 1 method)

At a particular point, is the function well-approximated?

f̃(0.1,0.2) ≈ f(0.1,0.2)

Does the inverse transform bring us back to the grid?

ipaduatransform(f̌) ≈ f̃.(x,y)

This page was generated using Literate.jl.